https://groups.google.com/forum/#!topic/google-appengine-downtime-notify/T_e7lVg7QNY
Wow, they have to take an entire datacentre offline, and they still don’t know what went wrong. (This was Tuesday, 25 June 2013. So, not Friday’s Google-wide outage, which remains unexplained, I think.)
Taking an entire DC offline is not terribly unusual at Google. But I can sympathize with not being able to find a root cause for a failure 